Harmonic Dataset Distillation for Time Series Forecasting

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Hong, Seungha, Jang, Sanghwan, Kweon, Wonbin, Kim, Suyeon, Lee, Gyuseok, Yu, Hwanjo
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866914368097091584
author Hong, Seungha
Jang, Sanghwan
Kweon, Wonbin
Kim, Suyeon
Lee, Gyuseok
Yu, Hwanjo
author_facet Hong, Seungha
Jang, Sanghwan
Kweon, Wonbin
Kim, Suyeon
Lee, Gyuseok
Yu, Hwanjo
contents Time Series forecasting (TSF) in the modern era faces significant computational and storage cost challenges due to the massive scale of real-world data. Dataset Distillation (DD), a paradigm that synthesizes a small, compact dataset to achieve training performance comparable to that of the original dataset, has emerged as a promising solution. However, conventional DD methods are not tailored for time series and suffer from architectural overfitting and limited scalability. To address these issues, we propose Harmonic Dataset Distillation for Time Series Forecasting (HDT). HDT decomposes the time series into its sinusoidal basis through the FFT and aligns the core periodic structure by Harmonic Matching. Since this process operates in the frequency domain, all updates during distillation are applied globally without disrupting temporal dependencies of time series. Extensive experiments demonstrate that HDT achieves strong cross-architecture generalization and scalability, validating its practicality for large-scale, real-world applications.
format Preprint
id arxiv_https___arxiv_org_abs_2603_03760
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Harmonic Dataset Distillation for Time Series Forecasting
Hong, Seungha
Jang, Sanghwan
Kweon, Wonbin
Kim, Suyeon
Lee, Gyuseok
Yu, Hwanjo
Machine Learning
Time Series forecasting (TSF) in the modern era faces significant computational and storage cost challenges due to the massive scale of real-world data. Dataset Distillation (DD), a paradigm that synthesizes a small, compact dataset to achieve training performance comparable to that of the original dataset, has emerged as a promising solution. However, conventional DD methods are not tailored for time series and suffer from architectural overfitting and limited scalability. To address these issues, we propose Harmonic Dataset Distillation for Time Series Forecasting (HDT). HDT decomposes the time series into its sinusoidal basis through the FFT and aligns the core periodic structure by Harmonic Matching. Since this process operates in the frequency domain, all updates during distillation are applied globally without disrupting temporal dependencies of time series. Extensive experiments demonstrate that HDT achieves strong cross-architecture generalization and scalability, validating its practicality for large-scale, real-world applications.
title Harmonic Dataset Distillation for Time Series Forecasting
topic Machine Learning
url https://arxiv.org/abs/2603.03760